Experiments and Field Studies with Insects on Vicia Sepium

Home , Bruchus, Cydia nigricana, Lathyrus pratensis, Triaspis (wasp), Vicia

Oecologia (2000) 122:129Ð137 © Springer-Verlag 2000

Andreas Kruess á Teja Tscharntke Species richness and parasitism in a fragmented landscape: experiments and field studies with insects on Vicia sepium

Received: 4 January 1999 / Accepted: 8 September 1999

Abstract Effects of habitat fragmentation on species di- Introduction versity and herbivore-parasitoid interactions were analyzed using the insect community of seed feeders and Habitat fragmentation following increasing intensity of their parasitoids in the pods of the bush vetch (Vicia se- land use in the landscape (Burgess and Sharpe 1981) has pium L.). Field studies were carried out on 18 old mead- been perceived as a major threat to biological diversity ows differing in area and isolation. The area of these (Wilcove et al. 1986; Noss 1991; Saunders et al. 1991; meadows was found to be the major determinant of spe- Tscharntke 1992; Rosenzweig 1995). The effects of hab- cies diversity and population abundance of endophagous itat fragmentation on species diversity can be mainly as- insects. Effects of isolation were further analyzed experi- signed to three processes: reduction of total habitat area mentally using 16 small plots with potted vetch plants within a region, loss of area within each single habitat, isolated by 100Ð500 m from vetch populations on large and increase in isolation between habitats (Andrén old meadows. The results showed that colonization suc- 1994). These three processes are interrelated in a non- cess greatly decreased with increasing isolation. In both linear way (Gustavson and Parker 1992). Theoretical cases, insect species were not equally affected. Parasito- models such as the equilibrium theory of island biogeo- ids suffered more from habitat loss and isolation than graphy (MacArthur and Wilson 1967) predict the number their phytophagous hosts. Minimum area requirements, of species on islands as a function of island size and iso- calculated from logistic regressions, were higher for lation. Loss of species may lead to changes in ecosystem parasitoids than for herbivores. In addition, percent para- functions such as decomposition, pollination, parasitism, sitism of the herbivores significantly decreased with area or predation (Kareiva 1987, 1990; Klein 1989; Kruess loss and increasing isolation of Vicia sepium plots, sup- and Tscharntke 1994, 1999; Didham et al. 1996; Burkey porting the trophic-level hypothesis of island biogeogra- 1997; Steffan-Dewenter and Tscharntke 1997; Didham phy. Species with high rates of absence on meadows and 1998; Dubbert et al. 1998; Tscharntke and Kruess 1999). isolated plant plots were not only characterized by their Ecosystem functions based on trophic structure were the- high trophic level, but also by low abundance and high oretically analyzed by Holt (1996), who investigated the spatial population variability. Thus conservation of large general effects of area loss and isolation on food chain and less isolated habitat remnants enhances species di- length and species of different trophic rank. He predicted versity and parasitism of potential pest insects, i.e., the that (1) food chain length should increase with area and stability of ecosystem functions. decrease with isolation, and (2) that the slope of species- area and species-isolation curves should increase with Key words Island biogeography á Insect diversity á trophic rank, either due to direct effects (e.g., decline of Herbivore-parasitoid interactions á Trophic levels á population size with trophic rank) or indirect effects (de- Biological control pendence of high-rank species on the presence of low- rank species). Not only trophic position but also feeding type may be differentially influenced by fragmentation, and several studies have examined the effects of fragmentation on both ectophagous (e.g., Davis 1975; Ward and Lakhani 1977; Rigby and Lawton 1981; Tscharntke 1992; A. Kruess (✉) á T. Tscharntke Matter 1997; Zabel and Tscharntke 1998) and endophag- Agroecology, University of Göttingen, Waldweg 26, D-37073 Göttingen, Germany ous insect communities on plants (MacGarvin 1982; e-mail: [email protected] Davis and Jones 1986; Tscharntke 1992; Kruess 1996; Tel.: +49-551-392359, Fax: +49-551-398806 Dubbert et al. 1998). 130 Table 1 List of landscape, meadow and host plant charac- Field studies on meadows Field experiment with Vicia sepium plots teristics used in multiple regression analyses Landscape characteristics Landscape characteristics Number of meadows within a radius of 250 m Total area of the “source meadow” (m2) Number of meadows within a radius of 500 m Distance to the source meadow (m) Total area of meadows within a radius of 250 m (m2) Distance to the nearest near-natural habitatc (m) Total area of meadows within a radius of 500 m (m2) Distance to the nearest fallow (m) Distance to the nearest meadow (m) Distance to the nearest hedge (m) Distance to the nearest meadow (>2 ha) (m) V. sepium plants mÐ2 on source meadows V. sepium pods mÐ2 on source meadows Meadow characteristics Meadow size (m2) Plant plot characteristics Center/edge ratioa Mean plant height (cm) Number of vascular plant species per 49 m2 Number of plants mÐ2 Cover of vegetation (%)b Number of pods mÐ2 Mean vegetation height (cm) Mean pod length (cm) Mean number seeds podÐ1 Host plant characteristics Number of legume species a Edge was defined as the 5 m Cover of legume species (%) wide margin. Center/edge ratio Number of Vicia species was calculated from the center Cover of Vicia species (%) area (m2) and the edge area Cover of V. sepium (%) (m2) Pods mÐ2 of V. sepium b Estimated by eye Mean pod length of V. sepium (cm) c Including hedges, forest Mean seeds podÐ1 of V. sepium edges, and fallows

We analyzed the effects of fragmentation on the endo- We addressed the following questions: phagous insects inhabiting the pods of bush vetch (Vicia 1. Are species richness and abundance affected by habi- sepium), including herbivores and parasitoids. Since en- tat area, habitat isolation, or other landscape, mead- dophagous insect communities are more likely to com- ow, or host plant characteristics? prise specialized mono- or oligophagous species than 2. Are parasitoids more affected than herbivores? ectophagous insects (Cornell 1989), area and isolation of 3. Does higher susceptibility of parasitoid species lead habitats can be defined by the distribution of the host to a reduction in parasitism? plant populations. Moreover, plant-insect systems with 4. How can species that are susceptible to habitat frag- endophagous species seem to be more likely to show mentation be characterized? changes in diversity and abundance due to environmental changes since they are more host-specific and less mobile (Cornell 1989). Previous experimental studies on the effects of isola- Material and methods tion on colonization success of endophagous insect her- Field samples were taken in a region called Kraichgau, situated bivores and their insect parasitoids on red clover (Kruess near Karlsruhe in south-west Germany. In the first half of this cen- and Tscharntke 1994, 1999; Kruess 1996) found that tury, this landscape was dominated by extensively used meadows food-web interactions were disrupted: on isolated plant with scattered standard cherry and apple trees. Most of these patches, parasitoid populations were more reduced than meadows had areas of several hectares, often surrounding villages like “green belts”. Intensification of agricultural land use since the their phytophagous hosts. In this paper, results are pre- 1930s has led to a dramatic loss and fragmentation of these mead- sented from field samples and from a field experiment ows. Nowadays, most of the remaining meadows are only small on the endophagous insect community in the pods of an- fragments of less than 1 ha. Few meadows of more than 10 ha still other legume, the bush vetch (Vicia sepium). Bush vetch exist. Within a region of 400 km2, 18 meadows with areas ranging 2 is a very abundant and widely distributed plant species in from 300 m to 70 ha were selected to analyze the effect of habitat area. Selection criteria were similarity in age, management, vege- old meadows in south-west Germany. The pods are easi- tation, exposition, and soil type (Loess). All meadows were at ly collected and rearing success of the insects is high. least 30 years old and were mown annually in mid-July. They Regional data on the range of host plants and hosts of the were scattered with apple trees and slightly exposed to the south expected insects were available from Garbe (1996), who or south-east. In June, vegetation of the meadows was mapped in 49-m2 plots (1×49 m2 on meadows <1000 m2, 2×49 m2 on mead- investigated the endophagous herbivores inhabiting the ows of 1000Ð10,000 m2, and 4×49 m2 on meadows >10,000 m2). pods of 27 leguminous plant species in the Kraichgau re- The parameters used to characterize the landscape surrounding the gion. Field samples were taken on extensively managed meadows, the meadows and the host plants and considered in the old meadows, to analyze the effects of habitat character- statistical analyses are shown in Table 1. From 3 to 6 July, 200 pods of the bush vetch V. sepium were collected randomly from istics (e.g., area, vegetation, surrounding landscape). The each meadow. The field experiment comprised 18 plots of potted field experiment tested the effects of isolation using V. sepium plants, each consisting of 12 pots, covering an area of small and isolated plots of potted vetches. 1.0 m2. Since this experiment was carried out to analyze the ef- 131 fects of isolation on colonization processes, we used five old Insect community meadows (area >2 ha) with naturally growing V. sepium as “sourc- es”. One plant plot was placed in the center of each meadow (con- trols), and the other 13 plant plots were established in the same lo- The endophagous insects found in the pods of V. sepium cal environment (old field margins adjacent to cereal fields) in the comprised 4 phytophagous and 10 parasitoid species area surrounding the meadows, separated by 100Ð500 m from the (Table 2). In the meadows, the herbivores were, from meadows. To avoid effects of plant variety, and nutrient and water highest to lowest abundance, the weevil Oxystoma availability, we used the same variety of V. sepium and the same potting compost for all plant plots. All plots were watered twice a ochropus, the seed-beetle Bruchus atomarius, the weevil week. Spatial arrangement of the plots was identical to that of an Tychius quinquepunctatus and the tortricid moth Cydia experiment done in 1992 with red clover plants (see Kruess and nigricana. The most abundant parasitoids were a pter- Tscharntke 1994 for details), but in June, two of the isolated plots omalid wasp Pteromalus sequester, two braconid wasps were destroyed, so results are based on only 16 Vicia plots. Pods Pigeria piger and Triaspis thoracicus, and an eupelmid were collected six times between 7 June and 7 July, since only a few ripened pods were available at the same time. All together wasp Eupelmus vesicularis. All other parasitoids had low 2032 pods were collected, with the total number of pods per plot abundances (≤1 individuals per 100 m2) and were found ranging from 101 to 155. Out of these, 100 pods per plot were ran- on only one or two meadows. domly selected for dissection, in order to have equal sample sizes. Five parasitoid species attacked the tortricid moth C. Abundances (e.g., specimens mÐ2, pods mÐ2) were calculated on a total sampled area of 6 m2 per plot (six sample times, plot size nigricana, four species parasitized the weevil O. ochro- 1 m2). pus, and one species, T. thoracicus, parasitized the seed- The pods from both the meadow samples and the field experi- beetle B. atomarius. The weevil Tychius quinquepuncta- ment were individually separated in small plastic tubes for rearing tus was not attacked by parasitoids (Table 2). In the field insects. After 5 months, the pods were dissected, and pod length and number of seeds were measured. Species number and species experiment, the number of insect species that successful- composition of herbivores and parasitoids were recorded. Un- ly colonized the small plots of potted V. sepium plants known insect species were sent to experts for identification (see was lower because there were feweer parasitoids: only Acknowledgements). four parasitoids were found on the control plots (see Statistical analyses (simple and multiple regression analyses, Table 2). logistic regression analyses, and one-way ANOVA) were per- formed with Statgraphics Plus for Windows 3.0 (Manugistics 1996). Data were tested for normality, and if necessary log-transformed. Percentages were always arcsine-transformed. Species diversity

The distribution of the insect species among the 18 dif- Results ferent-sized meadows is shown in Table 2. Three of the herbivores (O. ochropus, B. atomarius, and C. nigric- Meadow vegetation ana) were recorded on meadows of all size classes while the weevil T. quinquepunctatus was absent from mead- Altogether the 18 meadows supported 126 plant species ows smaller than 0.3 ha. This absence on small meadows that occurred in at least one meadow). Species diversity was significantly different from an equal distribution of ranged between 14 and 61 species of vascular plants per this species on all meadows (χ2=6.33, P=0.05). In con- 49 m2. Multiple regression showed that plant species trast to their hosts, the parasitoids were more scattered. richness was positively correlated with the meadow area Species diversity of parasitoids was highest on both the for both total number of vascular plant species largest and the smallest meadows, but species composi- [y=0.86+3.3×ln(x), F=20.4, P=0.003, r2=0.56, n=18] and tion differed. Three parasitoids were found only on large number of leguminous plant species [y=Ð0.2+0.61×ln(x), meadows (Scambus annulatus, Glabrobracon sp. 1, F=13.23, P=0.002, r2=0.45, n=18]. Abundance (% plant Entedon cf. procioni) or the smallest meadows cover) of V. sepium increased slightly from small to (Pristomerus vulnerator, Glabrobracon sp. 2, Tricho- large meadows [ln(y)=2.26+0.025×ln(x), F=5.0, r2=0.24, malus repandus), respectively. P=0.04, n=18], but neither pod (pods mÐ2) nor seed Regression analyses showed that species diversity of abundance (seeds mÐ2) was correlated with meadow area both herbivores and parasitoids increased with habitat ar- (pods: F=1.88, r2=0.19, P=0.19; seeds: F=0.39, r2=0.02, ea (Fig. 1a,b), but the slope of the species-area relation- P=0.54) or other habitat characteristics. Thus, resource ship was steeper for parasitoids than for herbivores. Only abundance for seed-feeding herbivores did not change two of the small meadows (<1 ha) supported more significantly from small to large meadows. than two parasitoid species, whereas in large meadows In the isolation experiment, seed set was negatively cor- (>1.6 ha) at least four parasitoid species were always related with isolation of the plots, slightly decreasing from found. Nevertheless, even in each of the large meadows, 3.8 seeds podÐ1 on the meadow plots to 3.3 seeds podÐ1 on species richness of parasitoids was only 50% of the total the most isolated plots (y=3.81Ð9.9×10Ð4x, F=10.3, richness of parasitoids. In the field experiment, multiple r2=0.43, P=0.006, n=16). However, the abundance of pods regression analyses showed a significantly negative spe- (pods mÐ2) and seeds(seeds mÐ2) were not negatively af- cies-isolation relationship for both herbivores and paras- fected by isolation (pods: F=0.07, r2=0.005, P=0.79; seeds: itoids. Number of herbivores decreased from four spe- F=1.48, r2=0.09, P=0.24). Thus, resource availability did cies on the control plots to two species on the most iso- not differ between isolated and non-isolated plant plots. lated plots (Fig. 1c). Number of parasitoids was approxi- 132 =16) n =11 ( =11 n on isolated isolated (total) 2 (min./max.) x Ð on control 100m (min./max.), plots with plots with A. glycophyllos, Lathyrus aphaca, L. pratensis, Lotus 2 x Ð =5 populations populations 100 m n Vicia cracca Vicia and =18 n Attacking tortricid moths. Also attacking hosts in pods of A polyphagous ectoparasitoid with a broad range of hosts e f g corniculatus mum and maximum abundance on colonized plots are given for control isolated plots. The numbers of colonized plots are listed for isolated and all For range of host plants and hosts see footnotes =4 n ), mini- Ð x =6 n =2 Vicia tetrasperma Vicia n and . Hosts of the parasitoids are indi- =6 n , but very occasionally 2 Vicia sepium Vicia corresponding with herbivore numbers. Meadow Astragalus glycophyllos, Cytisus scoparius, Genista (min./max.), =18 Vicia cracca Vicia x on all meadows >10 ha >1 ha >0.1 ha >0.01 ha All meadows plots 363 (156/709) 6128.7 (0/665) 6 2 6 1 3 4 Ð 18 10 576 (450/783) 225 (67/383) 450 (300/850) Ð 4 (9) 0 (5) Meadow studySpecimensper 100 m Number of meadows with populations Specimens per Specimens per Number of Field experiment Ð n and Lathyrus pratensis a b (Pz.) 1.0 (0/3)(Pz.) 1.0 Ð Ð Ð 1 1 Ð Ð g,4 g,4 only (Walk.) 0.13 (0/2) 0.13 (Walk.) Ð Ð Ð 1 1 Ð Ð Ð (L.) 261 (26/1286) 6 2 6 4 18 377 (250/700) 312 (66/517) (16) 11 (Walk.) (Walk.) 228.3 (0/516) 6 2 6 3 17 293 (217/383) 33 (33/33) 1 (6) (Retz.) 9.1 (0/36) 6 2 2 1 11 10 (0/17) Ð 0 (3) Lathyrus aphaca, L. pratensis, niger, d,1 1 (Erd.) 0.05 (0/0.7)1(Erd.) 0.05 Ð Ð Ð 1 Ð Ð Ð e,1 1 (Kiss) 1.0 (0/18)1(Kiss) 1.0 Ð Ð Ð 1 Ð Ð 2 f,4 4 4 (Curt) 16.1 (0/81) 6 1 Ð 1 8 3.3 (0/17) Ð 0 (3) (F.) (F.) 81.7 (0/397) 6 2 6 3 17 127 (100/167) (16/300) 112 (16) 11 c a ), minimum and maximum abundance, the number of meadows dif- sp. 1 sp. 2 4.2 (0/49) 1 0.3 (0/6) 1 Ð Ð Ð Ð Ð 2 1 1 Ð Ð Ð Ð (Wesm.) (Wesm.) 21.3 (0/190) 6 2 Ð 1 9 90 (0/150) Ð 0 (4) x Ð V. sepium V. procioni superscript arabic numerals The insect community in pods of cf. Oxystoma ochropus Bruchus atomarius quinquepunctatus Tychius Cydia nigricana Also attacking hosts in the pods of Also feeding on seeds of Feeding on Feeding on seeds of tinctoria, Lathyrus pratensis (Col., Bruchidae) (Hym., Braconidae) (Hym., Braconidae) Phytophagous insects 1. (Germ.) (Col., Apionidae) 2. 3. (L.) (Col., Curculionidae) 4. (Lep., Tortricidae) Parasitoids sequester Pteromalus (Hym., Pteromalidae) Eupelmus vesicularis (Hym., Eupelmidae) repandus Trichomalus Entedon thoracicus Triaspis Pigeria piger (Hym., Braconidae) Glabrobracon Scambus annulatus (Hym., Ichneumonidae) Pristomerus vulnerator Glabrobracon a b c d (Hym., Pteromalidae) (Hym., Braconidae) (Hym., Ichneumonidae) Range of hosts or host plants (Peck et al. 1964; Horstmann 1990; Garbe 1996) (Hym., Eupelmidae) Table 2 Table cated by study: mean ( ferent size classes with insect populations are given. Field experiment: mean ( 133

Fig. 1A,B Dependence of species richness on habitat size of dence of species richness on isolation of small Vicia plots._ old meadows. A Number of phytophagous insect species: C Number of phytophagous insect species: y=3.97Ð0.10√x, y=5.19Ð14.86/ln(x), F=28.7, r2=0.64, P<0.001, n=18. B Number F=179.3, r2=0.93, _P<0.001, n=16. D Number of parasitoid spe- of parasitoid species: y=Ð1.00+0.41/ln(x), F=17.7, r2=0.53, cies: y=2.31Ð0.13√x, F=52.13, r2=0.79, P<0.001, n=16. Compari- P<0.001, n=18. Comparison of regression lines of phytophagous son of regression lines of phytophagous and parasitoid species and parasitoid species showed significant differences in both slope showed no difference in the slopes but significant differences in (F=5.4, P=0.026) and intercept (F=4.8, P=0.037). C,D Depen- the intercepts (F=152, P<0.001)

Table 3 Results from logistic a χ2 regression analyses on the rela- Species abPPD Minimum area tion between species presence (m2) and meadow area [y=ea+bx/(1+ea+bx), x=ln(mead- Cydia nigricana Ð14.7119 2.4691 0.0185 0.73 5.5 942 ow area), y=species presence]. Tychius quinquepunctatus Ð8.98899 1.0322 0.0004 0.52 12.7 50881 Estimated minimum areas re- Eupelmus vesicularis Ð8.56803 1.0428 0.0007 0.48 11.5 30437 quired for 90% occupancy of Triaspis thoracicus Ð5.50586 0.5819 0.0056 0.31 7.7 561190 meadows for insect herbivores Pigeria piger Ð6.27436 0.6233 0.0034 0.35 8.6 799300 and parasitoids are given a Percentage of deviance (similar to r2 statistic) mately 2.5 species on the control plots and decreased to cupancy of habitats at only 942 m2. The polyphagous no species on the isolated plots (Fig. 1d). parasitoid E. vesicularis required 3 ha for 90% probabili- The presence-absence pattern of the insect species on ty of site occupancy, whereas its host O. ochropus colo- the meadows was analyzed by logistic regression. Both nized all meadows. Host abundance did not significantly ubiquitous species (Bruchus atomarius, Oxystoma ochro- contribute to the explanation of the presence-absence pus, Pteromalus sequester) and rare species (Table 2) pattern of these parasitoids in logistic regression. were excluded from analysis. For the two remaining herbivores (Cydia nigricana, Tychius quinquepunctatus) and the three parasitoids (Eupelmus vesicularis, Pigeria Species abundance and population variability piger, Triaspis thoracicus), the best-fit regression was obtained using log area (Table 3). Using the equations Only for one herbivore species was abundance related to from the logistic regressions, the minimum area require- habitat area. Number of specimens mÐ2 of Tychius quin- ment for a 90% probability of site occupancy was calcu- quepunctatus was positively correlated with meadow ar- lated (Table 3). In pairwise comparison, area require- ea [y=Ð1.12+0.46/ln(x), F=13.2, r2=0.45, P<0.01, n=18]. ments were higher for parasitoids than for their hosts. Host plant abundance (plant cover, pods mÐ2, seeds mÐ2) For example in T. thoracicus, the calculated minimum had no effect. The other herbivore species did not show area for 90% meadow occupancy was 56 ha, but its host, any correlation between abundance and habitat charac- B. atomarius, was distributed from the largest to the teristics (area, isolation) or host plant abundance. In the smallest meadows. The area requirement of P. piger was field experiment, abundances of two herbivores were highest at 80 ha, but its host C. nigricana had a 90% oc- found to be negatively affected by isolation. Density of 134 the weevil O. ochropus decreased from 525 specimens sults. Multiple regression analyses showed that absence 2 per 100 m on the control plots to 225 specimens per 100Ð rate was significantly correlated only with the coefficient m2 on plots isolated by 100Ð150 m (y=5.55Ð0.29√x, of variation (CV, see Fig. 2), but the coefficient of varia- F=67.5, r2=0.83, P<0.001, n=16, see also Table 2). More tion and the mean species abundance were negatively isolated plots were not successfully colonized. Abun- correlated [y=335.86Ð45.72 ln(x), F=107.4, r2=0.90, dance of the weevil T. quinquepunctatus was affected by P<0.001, n=14]. Thus, species that showed a high ab- isolation in that this species did not colonize any of the sence rate, were characterized by both low abundance isolated plant plots, though density of this weevil (488 and higher population variability. Moreover, species with specimens per 100 m2) on the control plots was very a high absence rate on the meadows also showed a high similar to that of O. ochropus. absence rate on the small plant plots and vice versa Two additional variables, the absence rate, and the (Spearman’s rank correlation: rs=0.8, P=0.034, n=8, spe- population variability, were calculated to characterize the cies found in the isolation experiment). species. Absence rate of a species was defined as percentage of the 18 meadows on which this species was not found. Population variability was calculated as coef- Parasitism ficient of variation (CV=100 s/Ðx) of species abundance on all 18 meadows. We used CV since it is insensitive Changes in percent parasitism were also analyzed with re- against differences between means and low abundances gard to habitat fragmentation as well as plant and host (Sokal and Rohlf 1995). We also included different cal- characteristics. For the three parasitized herbivores per- culations of CV (including or excluding zeros) in the re- cent parasitism on the meadows was only affected by hab- gression analyses but did not find differences in the re- itat size (Fig. 3a,b,c), whereas habitat isolation, host abundance, and plant characteristics had no significant effects. The weevil O. ochropus suffered most from parasitism, with a parasitism rate between 60% and 80% on large meadows, compared to 40% on the smallest meadows (Fig. 3a). Percent parasitism of the seed beetle Bruchus atomarius decreased from 25% on the largest meadows to nearly zero on the smallest meadows (Fig. 3b). Similarly, parasitism of the pea moth decreased from more than 20% on large meadows to nearly zero on the smallest meadows (Fig. 3c). In the field experiment, parasitism of the weevil O. ochropus was negatively affected by isolation (Fig. 3d). Though four isolated (by 100Ð150 m) Vicia plots Fig. 2 Percent absence of the herbivore (open circles) and parasi- were successfully colonized by O. ochropus (see Table 2), toid (filled circles) species on the meadows_ depending on popula- parasitoids attacked this weevil on only one plot. Percent tion variability (Cv): y=Ð40.01+6.17√x, F=102.5, r2=0.89, P<0.001, parasitism on this isolated plot was very low (19%) com- n=14 pared to parasitism on the control plots (53%).

Fig. 3AÐC Dependence of percent parasitism on habitat size of old meadows. A Oxystoma ochropus: y=56.03+5.64/ln(x), F=16.8, r2=0.51, P<0.001, n=18. B Bruchus atomarius: y=Ð17.51+2.79 ln(x), F=16.6, r2=0.51, P<0.001. C Cydia nigricana: y=Ð22.71+3.49 ln(x), F=7.3, r2=0.32, P=0.01, n=18. D Dependence of percent parasitism on isolation of small Vicia plots. Oxystoma_ ochropus: y=46.53Ð2.68√x, F=62.72, r2=0.82, P<0.001, n=16 135 Discussion characterized by small and variable populations, were more affected by fragmentation than species on the sec- The hypothesis that habitat area is an important determi- ond trophic level. This has been theoretically predicted nant of species diversity (Williams 1964; MacArthur and (Pimm 1991; Lawton 1995; Holt 1996) and empirically Wilson 1967; Brown 1971; Wilson and Willis 1975; supported by field studies (Kareiva 1987, 1990; Kruess Diamond and May 1976; Wilcox 1980; Wilcox and and Tscharntke 1994, 1999; Lei and Hanski 1997; Murphy 1985; Thornton et al. 1993) was supported by Roland and Taylor 1997; Zabel and Tscharntke 1998; our studies. These species-area effects may be explained Dubbert et al. 1998). In addition to the trophic-level po- by the area-per-se hypothesis or the habitat-heterogene- sition, parasitoids were less abundant and had more vari- ity hypothesis. In general, area-per-se effects can be able populations, which is generally found to be a major caused by: (1) a higher random extinction probability characteristic of species prone to extinction. As shown in due to smaller population sizes in smaller areas (Shaffer the field experiment with isolated Vicia plots, large in- 1981; Gilpin and Soulé 1986; Have 1993; Baur and sect populations on large meadows functioned as “source Erhardt 1995; Lei and Hanski 1997), or (2) the random- populations”, whereas less abundant species failed to sample effect in that larger samples have a higher proba- colonize plant plots isolated by 100 m or more (see also bility of containing more species (Connor and McCoy Kruess and Tscharntke 1994; Tscharntke and Kruess 1979; Haila 1983; Haila and Järvinen 1983). According 1999). Percent parasitism of the weevil O. ochropus was to the habitat-heterogeneity hypothesis species richness an increasing function of area, doubling from 35% on the is positively linked to habitat heterogeneity. On larger ar- smallest to 70% on the largest meadows. A similar pat- eas different habitat types are more likely to be included, tern was found for parasitism of two other herbivores, and different habitats support different communities, the seed-beetle B. atomarius, and the pea moth C. nigric- thus supporting higher species richness (Williams 1964; ana. This disruption in food-web interactions can be ex- Lack 1969, 1976; Johnson and Simberloff 1974; plained by a decline of both parasitoid species richness Williamson 1981; Rosenzweig 1995). Sample effects can and abundance of parasitoids caused by reductions in be excluded because the sample size was equal in all habitat area. Habitat isolation also negatively affected habitats. Since the herbivorous insects in this study are herbivore-parasitoid interactions, as shown in the field almost completely restricted to one plant species, the experiment: (1) on the isolated plant plots only one phy- bush vetch V. sepium, habitat heterogeneity should have tophagous species (O. ochropus) was attacked by paras- a low impact on species diversity of these herbivores. itoids compared to three species on the old meadows, (2) This is also supported by the results from multiple re- percent parasitism of O. ochropus was at least 50% low- gression analyses in that meadow characteristics (species er on the isolated plots than on the control plots and de- richness of vascular plants, legumes and Vicia species, creased to zero on the most isolated plots. The high sus- percent cover of total vegetation, legumes, and Vicia spe- ceptibility of the third trophic level to habitat fragmenta- cies) could not be related to species diversity. Further, re- tion supports previous results (Kruess and Tscharntke source concentration (Root 1973; e.g., pods mÐ2, and 1994, 1999; Zabel and Tscharntke 1998). These results seeds podÐ1) were independent of area. Since the field give support to the general conclusion that habitat frag- experiment (where landscape and habitat variables were mentation in the agricultural landscape releases herbi- kept constant) gave similar results to the meadow analys- vores from parasitism. Since pest populations are often es, we argue that possibly confounding landscape effects limited by top-down control of parasitoids (LaSalle and were of minor importance. Therefore, the species-area Gauld 1991; Hawkins and Gross 1992; LaSalle 1993), relationships found in this study are likely to be caused decreases in percent parasitism (shown in this study) by the high extinction probability of small populations. may sometimes favour pest outbreaks. This has been In fact, the probability of a species being absent was as- shown by Kareiva (1987) for an aphid-predator system sociated with small populations, but even more with spa- in which population explosions of aphids were more fre- tial population variability (as in Kruess and Tscharntke quent on small than large patches of goldenrod. Roland 1994). Since the small meadows were remnants of for- and Taylor (1997) found that outbreaks of the tent cater- mer large meadows, it is most likely that historically spe- pillar Malacosoma disstria lasted longer in fragmented cies richness of the former large meadows had been sim- forest due to low parasitism. In bush vetch V. sepium, ilar to species richness of the large meadows of this changes in herbivore abundance with area did not show a study. Thus decline in species richness on the small general pattern. Only in the weevil T. quinquepunctatus meadows can be attributed to extinction processes. Ac- were abundance and area correlated. But this species was cording to the results from our field experiment, only the not attacked by parasitoids and, opposite to the findings most abundant species colonized small meadows. De- of to Kareiva (1987), abundance was positively correlat- cline in species richness with decrease in habitat area ed with area. Since T. quinquepunctatus did not occur on was steeper for parasitoids than for their phytophagous meadows smaller than 3000 m2, it is likely that a critical hosts. In logistic regression analyses of the percent ab- threshold in minimum area requirement (Kareiva and sence of species and meadow area, minimum area re- Wennergren 1995; With and Christ 1995; Andrén 1996; quirements were up to 10 times higher for parasitoids Bascompte and Solé 1996) caused this abundance-area than for herbivores. Species on the third trophic level, pattern. 136 In conclusion, the results of this paper indicate that servation biology: the science of scarcity and diversity. Sinau- remnant large habitats play an important role in (1) the er, Sunderland, pp 13Ð34 Gustavson EJ, Parker GR (1992) Relationships between landcover conservation of species richness of insect communities, proportion and indices of landscape spatial pattern. Landscape and (2) the enhancement of parasitoid populations rela- Ecol 7:101Ð110 tive to herbivore populations, and thereby, in the preser- Haila Y (1983) Land birds on northern islands: a sampling meta- vation of herbivore-parasitoid interactions. Habitat isola- phor for insular colonization. Oikos 41:334Ð351 Haila Y, Järvinen O (1983) Land bird communities on a Finnish is- tion of only a few hundred meters negatively affected the land: species impoverishment and abundance patterns. Oikos 41: colonization process of insects and destabilized herbi- 255Ð273 vore-parasitoid interactions. Species that were most like- Have A (1993) Effects of area and patchiness on species richness: ly to be negatively affected by habitat fragmentation an experimental archipelago of ciliate microcosmos. Oikos 66: could be characterized by their trophic position, which 493Ð500 Hawkins BA, Gross P (1992) Species richness and population lim- was associated with small and variable populations. itation in insect parasitoid-host systems. Am Nat 139:417Ð423 Holt RD (1996) Food webs in space: an island biogeographical Acknowledgements Taxonomic help came from M. Čapek (Bracon- perspective. In: Polis GA, Winemiller KO (eds) Food webs. idae), K. Horstmann (Ichneumonidae), and S. Vidal (Chalcidoidea), Chapman Hall, New York, pp 313Ð323 and financial support from the Deutsche Forschungsgemeinschaft Horstmann K (1990) Die westpaläarktischen Arten der Gattung (DFG). Comments of Bradford A. Hawkins and two anonymous ref- Pristomerus Curtis, 1836 (Hymenoptera, Ichneumonidae). Z erees improved an earlier version of this paper. Entomol 11:9Ð44 Johnson MP, Simberloff DS (1974) Environmental determinants of island species numbers in the British Isles. Biogeography 1: References 149Ð154 Kareiva P (1987) Habitat-fragmentation and the stability of preda- Andrén H (1994) Effects of habitat fragmentation on birds and tor-prey interactions. Nature 326:28Ð290 mammals in landscapes with different proportions of suitable Kareiva P (1990) Population dynamics in spatially complex envi- habitat: a review. Oikos 71:355Ð366 ronments: theory and data. Philos Trans R Soc Lond B 330: Andrén H (1996) Population responses to habitat fragmentation: 175Ð190 statistical power and the random sample hypothesis. Oikos 76: Kareiva P, Wennergren U (1995) Connecting landscape patterns to 235Ð242 ecosystems and population processes. Nature 373:299Ð301 Bascompte J, Solé RV (1996) Habitat fragmentation and extinc- Klein BC (1989) Effects of forest fragmentation on dung and car- tion thresholds in spatially explicit models. J Anim Ecol 65: rion beetle communities in central Amazonia. Ecology 70: 465Ð473 1715Ð1725 Baur B, Erhardt A (1995) Habitat fragmentation and habitat alter- Kruess A (1996) Folgen der Lebensraum-Fragmentierung für ations: principal threats to most animal and plant species. Gaia Pflanze-Herbivor-Parasitoid-Gesellschaften: Artendiversität und 4:221Ð226 Interaktionen. Haupt, Bern Brown JH (1971) Species richness of boreal mammals living on Kruess A, Tscharntke T (1994) Habitat fragmentation, species the montane islands of the Great Basin. Am Nat 105:467Ð478 loss, and biological control. Science 264:1581Ð1584 Burgess RL, Sharpe DM (eds) (1981) Forest island dynamics in Kruess A, Tscharntke T (1999) Effects of habitat fragmentation on man-dominated landscapes. Springer, Berlin Heidelberg New plant-insect communities. In: Ekbom B, Irwin M, Robert Y York (eds) Interchanges of insects between agriculture and sur- Burkey TV (1997) Metapopulation extinction in fragmented land- rounding landscapes. Kluwer, Dordrecht, pp 53Ð71 scapes: using bacteria and protozoa communities as model Lack D (1969) The numbers of bird species on islands. Bird ecosystems. Am Nat 150:568Ð591 Study 16:193Ð203 Connor EF, McCoy ED (1979) The statistics and biology of the Lack D (1976) Island birds. Blackwell, Oxford species-area relationship. Am Nat 113:791Ð833 LaSalle J (1993) Parasitic Hymenoptera, biological control and Cornell HV (1989) Endophage-ectophage ratios and plant defense. biodiversity. In: LaSalle J, Gauld JD (eds) Hymenoptera and Evol Ecol 3:64Ð76 biodiversity. CAB International, Wallingford, pp 197Ð215 Davis BNK (1975) The colonization of isolated patches of nettles LaSalle J, Gauld JD (1991) Parasitic hymenoptera and biodiversi- (Urtica dioica L.) by insects. J Appl Ecol 12:1Ð14 ty crisis. Redia 74:315Ð334 Davis BNK, Jones PE (1986) Insects on isolated colonies of com- Lawton JH (1995) Population dynamic principles. In: Lawton JK, mon rock-rose Helianthemum chamaecistus. Ecol Entomol 11: May RM (eds) Extinction risks. Oxford University Press, Ox- 267Ð281 ford, pp 147Ð163 Diamond JM, May RM (1976) Island biogeography and the design Lei G-C, Hanski I (1997) Metapopulation structure of Cotesia me- of natural reserves. In: May RM (ed) Theoretical ecology. litaearum, a specialist parasitoid of the butterfly Melitaea cin- Blackwell, Oxford, pp 228Ð252 xia. Oikos 78:91Ð100 Didham RK (1998) Altered leaf-litter decomposition rates in trop- MacArthur RH, Wilson EO (1967) The theory of island biogeog- ical forest fragments. Oecologia 116:397Ð406 raphy. Princeton University Press, Princeton Didham RK, Ghazoul J, Stork NE, Davis AJ (1996) Insects MacGarvin M (1982) Species-area relationships of insects on host in fragmented forests: a functional approach. Trends Ecol plants: herbivores on rosebay willowherb. J Anim Ecol 51: Evol 11:255Ð260 207Ð223 Dubbert M, Tscharntke T, Vidal S (1998) Stem-boring insects of Manugistics (1996) Statgraphics+ for Windows, version 2, user fragmented Calamagrostis habitats: herbivore-parasitoid com- manual. Manugistics, Rockville munity structure and the unpredictability of grass shoot abun- Matter SF (1997) Population density and area: the role of be- dance. Ecol Entomol 23:271Ð280 tween- and within-patch processes. Oecologia 110:533Ð538 Garbe P (1996) Phytophage und Parasitoide in Früchten der Fa- Noss RF (1991) Landscape connectivity: different functions at dif- baceae: Artenreichtum und -häufigkeit in Abhängigkeit von ferent scales. In: Hudson WE (ed) Landscape linkages and Wirtspflanzen- und Lebensraummerkmalen. Diploma Thesis, biodiversity. Island Press, Washington DC Institute of Zoology, University of Karlsruhe Peck O, Boucek Z, Hoffer A (1964) Keys to the Chalcidoidea of Gilpin ME, Soulé ME (1986) Minimum viable populations: the Czechoslovakia (Insecta: Hymenoptera). Mem Entomol Soc processes of the species extinctions. In: Soulé ME (ed) Con- Can 34:1Ð120 137 Pimm SL (1991) The balance of nature? Ecological issues in the Tscharntke T, Kruess A (1999) Habitat fragmentation and biologi- conservation of species and communities. The University of cal control. In: Hawkins BA, Cornell HV (eds) Theoretical ap- Chicago Press, Chicago proaches to biological control. Cambridge University Press, Rigby C, Lawton JH (1981) Species-area relationships of arthro- Cambridge, pp 190Ð205 pods on host plants: herbivores on bracken. J Biogeogr 8: Ward LK, Lakhani KH (1977) The conservation of Juniper: The 125Ð133 fauna of food-plant island sites in southern England. J Appl Roland J, Taylor PD (1997) Insect parasitoid species respond to Ecol 14:121Ð135 forest structure at different spatial scales. Nature 386:710Ð713 Wilcove DS, McLellan CH, Dobson AP (1986) Habitat fragmenta- Root RB (1973) Organization of a plant-arthropod association in tion in the temperate zone. In: Soulé ME (ed) Conservation bi- simple and diverse habitats: the fauna of collards, Brassica ol- ology. The science of scarcity and diversity. Sinauer, Sunder- eracea. Ecol Monogr 43:95Ð124 land, pp 237Ð256 Rosenzweig ML (1995) Species diversity in space and time. Uni- Wilcox BA (1980) Insular ecology and conservation. In: Soulé versity Press, Cambridge ME, Wilcox BA (eds) Conservation biology: an evolutionary- Saunders DA, Hobbs RJ, Margules CR (1991) Biological conse- ecological perspective. Sinauer, Sunderland, pp 95Ð117 quences of ecosystem fragmentation. Conserv Biol 5:18Ð32 Wilcox BA, Murphy DD (1985) Conservation strategy: the effects Shaffer ML (1981) Minimum population sizes for species conser- of fragmentation on extinction. Am Nat 125:879Ð887 vation. Bioscience 31:131Ð134 Williams CB (1964) Patterns in the balance of nature and related Sokal RR, Rohlf FJ (1995) Biometry, 3rd edn. Freeman, New problems in quantitative ecology. Academic Press, New York York Williamson M (1981) Island populations. Oxford University Press, Steffan-Dewenter I, Tscharntke T (1997) Bee diversity and seed Oxford set in fragmented habitats. Acta Hortic 437:231Ð234 Wilson EO, Willis EO (1975) Applied biogeography. In: Cody Thornton IWB, Zann RA, Balen S van (1993) Colonization of ML, Diamond JM (eds) Ecology and evolution of communi- Rakata (Krakatau Is.) by non-migrant land birds from 1883 ties. Harvard University Press, Cambridge, pp 522Ð534 to1992 and implications for the value of island equilibrium With KA, Christ TO (1995) Critical threshold in species responses theory. J Biogeogr 20:441Ð452 to landscape structure. Ecology 76:2446Ð2459 Tscharntke T (1992) Fragmentation of Phragmites habitats, mini- Zabel J, Tscharntke T (1998) Does fragmentation of Urtica habi- mum viable population size, habitat suitability, and local ex- tats affect phytophagous and predatory insects differentially? tinction of moths, midges, flies, aphids and birds. Conserv Oecologia 116:419Ð425 Biol 6:530Ð536